Fish Audio × SLNG Partnership

Fish Audio and SLNG have partnered to accelerate the global availability of Fish's latest voice models. Fish Audio S2.1 Pro is now live in the EU and Australia. Fish builds great models, while SLNG runs the infrastructure they sit on.

Most voice models today are served from a handful of US regions. The round trip to Jakarta or São Paulo comes out of the same latency budget the agent needs for everything else.

This partnership puts Fish's models on physical compute in the markets where they're used, under the residency rules those markets apply.

From Fish: the models.

Fish's latest real-time models set a high bar on first-byte latency. S2.1 Pro covers 83 languages, with top-tier Chinese, Japanese and English, and strong coverage across Korean, Spanish, Portuguese, Arabic, Russian, French, German, Hindi, Tamil, Thai and dozens more.

"Voice AI is global by nature, but delivering it well requires local infrastructure. Partnering with SLNG puts our latest models in more regions, for companies that want to ship production voice AI with the latency and the enterprise controls those markets require"

— Rissa Cao, CEO, Fish Audio

From SLNG: the infra.

SLNG runs global voice infrastructure. Fish's models are deployed into it directly, on physical GPU hardware in each region rather than proxied back to a US endpoint. That means a Fish model reaches a new market as fast as SLNG can stand up compute there, and it arrives with the region's controls already applied.

In-region processing is enforced at the routing layer. ISO 27001 certified, HIPAA and GDPR compliant. Audio never leaves the region it was processed in.

"Voice agents are one of the most latency-sensitive workloads anyone is building right now. The model has to sit close to the user. Fish is one of the strongest voice model families out there, particularly for real-time and for the languages most providers don't cover well. Bringing them into the EU and Australia is exactly the kind of pairing we set SLNG up to make possible."

— Luke Miller, CEO, SLNG

SLNG + Fish

Fish models plus the SLNG execution layer is a voice agent that runs start to finish inside one region. Speech in, model, speech out, all on the same regional compute. No ocean crossing per turn, and nothing to explain to a compliance team. Audio processed in the EU stays in the EU. Same in Australia. An agent can be scoped to a single country and still run at the cost and turn latency of a US deployment.

Read the docs.

Looking ahead

Jakarta, Mumbai, UAE, São Paulo, Seoul, Singapore and Saudi Arabia are lined up next.

If you're building voice agents and need Fish in a specific region, tell us which one.

slng.ai · fish.audio